Professional Documents
Culture Documents
ABSTRACT: In the field of reinforced-concrete (RC) structures, the use of fiber reinforced plastic rebars (FRP
rebars) as an alternative to the steel reinforcements appears very promising, especially if such structures are
exposed to corrosive environments. However, a better understanding of the mechanical behavior of FRP
reinforc~ments-in particular bond behavior-is needed in order to use them for practical purposes. For this
reason, In the last few years a number of tests on several types of FRP rebars has been conducted in order to
evaluate the interaction phenomena between FRP rebars and the concrete matrix and to evidence behavioral
differences with respect to the deformed steel rods. In this paper a state-of-the-art report on the bond of FRP
bars to concrete is presented. Numerous tests are analyzed to better understand bond mechanisms and the
influence of type of fiber, outer surface (shape and type of matrix), and other significant parameters (Le.,
confining pressure, bar diameter, compressive concrete strength) on bond performances. Furthermore, some
Downloaded from ascelibrary.org by University of Leeds on 07/01/18. Copyright ASCE. For personal use only; all rights reserved.
analytical models of bond-slip behavior are examined to assess their adequacy to reproduce the experimental
bond behavior. In particular, the investigation focuses on the reliability of the well-known model by Malvar (the
first one dedicated to FRP reinforcements) as well as on the model by Eligehausen, Popov, and Bertero, devel-
oped for .steel reinforcements but successfUlly applied to FRP ones. In addition, the effectiveness of two analytical
formulations proposed by the authors, the first one representing the ascending branch of the bond-slip curve and
the second the entire curve, is demonstrated.
bars and concrete has been investigated by many authors, with is a result of mechanical interlocking. As a confirmation, the
reference to several types of bars characterized by various sur- results reported by Nanni et ai. (1995) show that in this case
face configuration, by different quality and quantity of fibers, bond strength is not dependent on concrete strength; in fact,
and by the use of different resins as binders. From the ex- the concrete around the reinforcement is stressed to a small
perimental results it is possible to observe that the bond of degree by bond mechanisms and there are no microcracks.
FRP reinforcement to concrete is controlled by several factors, For what concerns smooth rebars, friction appears to be the
including the following: chemical bond; friction due to surface major source of pullout resistance since adhesion vanishes as
roughness of FRP rods; mechanical interlock of the FRP rods soon as the rod slips relative to the concrete. Moreover, the
against the concrete; hydrostatic pressure against the FRP rods experimental tests have shown that the friction component of
due to shrinkage of hardened concrete; and swelling of FRP bond strength essentially depends on the resin quality; in fact,
rods due to temperature change and moisture absorption. Dur- the different smoothness of the resin influences the bond.
ing initial pullout, the chemical bond (adhesion) is the main Two further effects influence pullout resistance: transverse
resisting mechanism; afterwards, it is replaced by friction and! hydrostatic strains that develop within the rod in addition to
or mechanical interlock. Since chemical bond between con- longitudinal strain during the debonding process, and the hy-
crete and FRP generally is extremely low, friction and me- drostatic pressure due to shrinkage of surrounding hardened
chanical interlock really become the primary mean of stress concrete. The last effect could become significant depending
transfer. on the specimen condition (AI-Zahrani 1995).
Based on these considerations, a first attempt at classifica- To verify that surface condition has a significant effect on
tion for bond mechanism may come from the work by Kan- bond resistance of smooth FRP rebars, AI-Zahrani studied the
akubo et ai. (1993). The authors identified two types of bond- experimental behavior of smooth glass-vinyl ester rods after
resistant mechanisms: one is the friction-resistant type, while removing the resin-rich layer. The experimental results showed
the other one is the bearing;.resistant type (when the mechan- an increase in the bond strength (five times higher than unal-
ical interlock becomes dominant). According to Kanakubo et tered smooth rods) due to the "skin-removed effect," proving
aI., smooth and strand-shaped bars exhibit a friction-resistant that the rod surface condition (composition of the resin-rich
bond while ribbed FRP bars develop a bearing-resistant bond. surface layer and its smoothness) clearly affects the adhesion
Another classification for bond behavior was suggested by and friction components of the bond strength.
Makitani et al. (1993). The authors defined two possible be- The influence of the smoothness of the outer surface on the
haviors depending on the adopted surface process: the first, bond behavior of straight FRP rebars is confirmed by the test
typical of straight grain-covered rebars, is characterized by a results obtained by Makitani et ai. (1993) for sanded rebars.
very small slip (sm) corresponding to the peak bond strength According to their results, sanding leads to an increase of
(Tm); the second, relative to braided rods, is characterized by chemical bond, thus resulting in a large increment of bond
much larger values of the slip Sm' strength. However, sand-covered continuous fiber bars show
It should be noticed, based on observations made by other good bond performance initially, but the interface between
researchers, that the preceding classifications are rather inad- sand grains and bars detaches abruptly, with a brittle bond
equate, especially when characterizing the mechanism of load failure (Itoh et ai. 1989).
transferring along the interface between FRP-deformed rods An intermediate behavior between smooth and sanded re-
and concrete. In this case, the anisotropic nature of the FRP bars has been found by Alunno Rossetti et al. (1995) and
materials needs to be accounted for since it results in different Faoro (1992) for rebars obtained with some different external
mechanical and physical properties in the longitudinal and surface process (i.e., sandblasted-type rebars, bars with epoxy
transverse directions. Anisotropy is a result of the fact that powder coating, bars with polyamide sheathing). Such tests
shear and transverse properties of bars are dominated by resin have proved that chemical bond is not negligible any more;
while longitudinal properties are influenced by fibers. Resin- however, bond resistance has resulted in values lower than
dependent strength may be lower than the concrete compres- those for grain-covered rods.
sive strength, thus resulting in a different bond interaction
from that of steel reinforcements with failure due to damage Deformed FRP Rods
of ribs instead of cracking of concrete. Therefore, failure
stresses and strains in the longitudinal and transverse direc- The previous sections emphasized that, in order to improve
tions are very influential on bond behavior. Such properties- bond behavior between concrete and FRP rebars, some pro-
highly dependent on temperature and environmental condi- cesses have been developed to produce deformed rods, thus
tions-are especially important to determine the pullout obtaining a wide range of commercial types of rebars. It is
strength and the earlier failure modes of FRP rods (matrix- or worth mentioning the following types of deformed FRP rods:
fiber-dominated failure modes). ribbed-type rods, braided rods, indented rods, twisted strands,
Although bond behavior is dependent on various effects, it spiral glued-type rebars; moreover, within each type, various
JOURNAL OF COMPOSITES FOR CONSTRUCTION / MAY 1997/41
deformation in a plane containing the radial and longitudinal rods); the second one had an external helicoidal tow that
directions must develop. The longitudinal shear modulus of provided a protruding deformation (ribbed type). In these
the FRP rod in this plane is the ruling constitutive parameter cases, the observation of specimens after tests has shown both
for such action. a certain damage of the outer surface of the bars and a wide
Therefore, bond behavior of the deformed rebars depends diffusion of cracks in the concrete matrix; such cracking evi-
on the manner the interlocking interaction possibly activates; denced that a bearing-type mechanism was activated.
this is influenced by the rebar geometry (spacing and dimen- A similar observation was reported by Hattori et al. (1995)
sion of ribs), the concrete properties, and mainly the mechan- in the case of spirally wounded aramid bars. In such case, the
ical characteristics of the reinforcement. authors noticed a dependence of the maximum bond stress on
In some conditions deformed rebars do not develop inter- the concrete compressive strength; they also found that the
locking mechanisms. As an example Malvar (1994) found that mechanical properties of binding material (resin), as well as
deformed rebars obtained by gluing a fiber spiral on the outer those of concrete, have great influence on the pullout behavior
surface possess a bond behavior similar to that of smooth re- of the FRP rebars. Therefore, the authors concluded that the
bars with no mechanical interlock. In this case, tests have pullout mechanism of the FRP reinforcements differ from that
shown that the bond collapse is due to the detachment of the of deformed steel bars, whose bond behavior is influenced by
spiral, whereas the concrete remains uncrushed. Therefore, strength of the concrete matrix only.
bond performances are independent of both concrete com- Differences between bond behavior of FRP and conven-
pressive strength and confinement pressure. tional deformed steel reinforcements were also noticed by
An analogous behavior has been found by Hattori et a1. Chaallal and Benmokrane (1993); they found that bond
(1995) for CFRP cables consisting of seven straight bundles strength of GFRP deformed rods embedded in high-strength
of carbon fibers that are twisted into a cable (twisted strands). concrete (HSC) is similar to that obtained when such rods
Observation of the interface after testing indicated that the were embedded in normal-strength concrete (NSC). Their find-
rod position in the specimen (top or bottom rebar), embedment authors are often rather small (on the average five to 10 times
length, temperature change, and environmental conditions. the bar diameter).
In the following, a critical review of an extensive set of
such bond tests is presented. For the sake of clarity, the ex-
perimental results are classified based on the shape of the outer Straight FRP Rods
surface as previously done for bond mechanisms. In particular, Tests on smooth as well as grain-covered FRP rods are con-
the average performances in terms of strength and deform- sidered here (see Table 1). It can be seen that, apart from test
ability are evaluated for the two categories of rebars (straight procedure and material mechanical properties, bond strength
and deformed rods). Within each category, the further depen- (Ton) for smooth rebars is rather poor (Ton < 2.37 MPa).
dence of bond on the mechanical properties of reinforcements A remarkable increase of Ton is obtained with proper surface
and concrete is also evidenced. The effect of the other afore- treatments. By sanding the outer surface of smooth rebars,
mentioned factors is discussed in the following section. bond resistance Ton greatly amplifies (Ton> 10 MPa), thus at-
Based on the proposed classification, the main data of the taining values similar or even larger than those of steel rein-
considered tests are reported in Tables 1 and 2 for straight and forcements.
deformed rods, respectively. The two tables report on: type of Table 1 also reports test data for the so-called sandblasted-
rod (i.e., smooth, ribbed, indented, braided); test identification; type rebars whose surface is treated differently (Alunno Ros-
type of used fiber; description of the type of outer surface and setti et al. 1995).
type of matrix; values of the maximum bond stress (Ton); and Although values of Ton are lower than those of sanded rebars,
references. a remarkable increase (Ton up to 5.41 MPa) is obtained com-
Part of the selected experimental data come from pullout pared to smooth rebars.
tests and the rest from hinged-beam tests. It should be em- Such increments in strength are due to an increase of chem-
phasized that a pullout test does not reflect the actual condition ical bond (up to 10 times for grain-covered rebars) and of the
of reinforcements in reinforced-concrete members under bend- friction coefficient associated with the outer-surface roughness.
ing (Larralde and Silva Rodriguez 1993); in fact, a compres- However, for grain-covered rebars the increases in bond
strength are counterbalanced by a more brittle T-S relationship.
TABLE 3. Mean Values and Coefficients of Variation for Test The mean values of the bond strength Ton and the associated
Results slip Sm are reported in Table 3 along with the coefficient of
Number 'T m sm variation (COV). For what concerns smooth rods, their relative
Type of outer surface of tests (MPa) COV (mm) COY low bond values (Ton = 1.19 MPa) lead to the conclusion that
(1 ) (2) (3) (4) (5) (6) it is not safe to use smooth bars as reinforcement in concrete
Smooth rebars 10 1.19 0.44 0.26 1.29 structures.
Grain covered 5 12.05 0.09 0.13 0.37 Fig. 1 shows the experimental values of Ton and Son' with their
Sandblasted-type rebars 10 2.74 0.52 1.08 1.73 means, for smooth and grain-covered rebars.
Twisted strands 3 4.15 0.08 0.45 0.24 For smooth rods, it has to be shown that the effect of con-
Ribbed type 27 11.61 0.34 1.23 1.19 crete mechanical properties appear to be negligible, being the
Indented and braided 12 10.20 0.10 2.14 0.61
2 17.78 0.08 0.15 0.00
bond performances dependent on the type of fibers and ma-
Braided and sanded
trix only. This has been found by Nanni et al. (1995), who
25 ~--------"'---------:-1 25 ~-----.-----'- - - - - - - - - - - - ,
m[MPa] Smooth 1: ",IMI-nt Grain covered
(a) 20
(b)
20
15
15
•
10 10
,-
····t······ •
A = mean value
A = mean value
5 5
Behavior
When sand coating is applied to deformed FRP rods, values
12...----------------------, of T m are even larger (>16 MPa), thus exceeding the bond
............ SmO(l/h
.•••••. SarulhIQS,-.J-/l~
strength of deformed steel rods. In fact, sand grains that are
- - Grain covered glued on the surface, provide a further increase in strength and
9 stiffness.
Fig. 4 shows values of T m and of the corresponding slip Sm
for all types of deformed rods. It can be seen that such values
6 are rather scattered, especially those for rib-type rods. Such
GFSA
scatter is due to the large variability of the tested rebars since
no standardization has been developed yet. Namely, the dif-
ferences in geometry and in the mechanical properties of fibers
/ ~~~. and resins, resulting (as already said) in the activation of dif-
.......................................·..· · · GV6 ferent mechanisms, generate rather different bond resistance.
The experimental T-S curves for some deformed rods are
2 s[mm] 4 shown in Fig. 5: a large variation in the bond resistance as
well as in the overall bond behavior emerges. Twisted strands
FIG. 3. Bond-5l1p Experimental Laws for Smooth, Graln-Cov-
ered, and Sandblasted-TYpe Rebars present a clearly frictional behavior with a constant plateau at
about 3 MPa.
carried out tests on smooth rods by varying the type of fiber From the experimental bond-slip curves for ribbed and in-
(carbon or glass) and matrix (epoxy or vinyl ester resin). dented rods, it can be concluded that the ultimate bond
Fig. 2 reports some of their results, which clearly show the strength Tm is somehow similar to that of steel reinforcements
dependence of T-S curves on matrix and fibers, confirming that but occurs at a considerably larger level of slip (sm)' Finally,
in the case of smooth rebars the composition of the resin-rich the T-S curves for sanded deformed rods evidence a noteworthy
surface layer and its smoothness are the critical parameters. increase in adhesion, a considerable bond resistance, a limited
All curves in Fig. 2, as well as those representing experi- slip Sm' and a strong slope of the softening branch.
mental T-S relationships in the subsequent figures, are labeled
with the same identifications reported in Tables I and 2 in EXPERIMENTAL RESULTS: EFFECT OF OTHER
order to enable definition of their source. PARAMETERS
Further observation about the T-S relationship can be drawn The large scatter in the data presented in previous sections
from Fig. 3, which displays typical T-S curves for smooth, is partially due to the influence of parameters in addition to
grain-covered, and sandblasted-type rebars. For smooth and those already considered. The most significant observations on
sandblasted-type rods, it can be seen that a rather constant the effect of t4ese further parameters are summarized in the
trend characterizes the curves after attaining T m (friction be- following section.
havior); furthermore, very low values of adhesion are ob-
served. For grain-covered rebars, the behavior is quite different
Effect of Confinement Pressure
since values of bond strength are larger, chemical bond is not
negligible, stiffness is higher, and then values of Sm are lower. The effect of confinement pressure on the interaction be-
In addition, the softening branch is fastly. degrading after tween reinforcing rebars and surrounding concrete has been
reaching the maximum bond strength, denoting a very brittle studied by Soroushian et al. (1991) in the case of steel rein-
behavior. forcements.
For what concerns FRP rebars, an extensive study was car-
Deformed FRP Rods
ried out by Malvar (1994), which was aimed at investigating
As already mentioned the surface treatments, leading to fab- the bond-slip behavior of four commercially available FRP
rication of deformed FRP rods, intend to increase friction, reinforcing rebars. The main characteristics of the selected bars
compared to smooth rods, and to obtain further contributions (identified as A, B, C, and D) are reported in Table 2. The
to bond strength from mechanical interlocking. experimental data were obtained for five levels of constant
Several types of rebars belong to this category, having dif- radial-confining stress. The obtained results have shown a
ferent bond mechanisms, which result in different bond-resis- strong dependence of bond strength on the confinement for
tance and bond-constitutive relationships. This can be seen both deformed and stressed rebars (types A, B, and D), while,
from Table 2, which reports values of Tm for each considered for bars with a spiral glued to the outer surface (type C), the
test, and from Table 3, which reports average values of Tm and response was almost unaffected by confinement. In this last
Sm over the entire experimental test set. case, in fact, it was observed that the spiral (initially glued to
JOURNAL OF COMPOSITES FOR CONSTRUCTION f MAY 1997/45
15 15
• .: •
10
6. = mean value
10 It· . &. .., .... •
.....................................................................
6. =mean value
I
!
••
5 ,
4· ...................... 5 •
i
i Sill (Ilun)
• _Il.[mln)
0
0.1 0.2 0.3 0.4 0.5 0.6 0 2 4 6 8 10
25
.r
Downloaded from ascelibrary.org by University of Leeds on 07/01/18. Copyright ASCE. For personal use only; all rights reserved.
Ribbed
(c)
i Deformed & Sanded
20 (~
.....................................+. .
• 15
i 6.= mean value
10 i
5 •
• 6. = mean value
5 I
8 .. [mml 1 8 m [mml
00·1---1..-.-'--1..-.--1..-.--1..-.--1..-.----..1
2 4 6 8 10 0.1 0.2 0.3 0.4 0.5 0.6
bon, aramid, glass, and vinylon rebars-characterized by dif- bond-slip constitutive law is needed. In spite of the numerous
ferent shapes of the outer surface (smooth, grain-covered, formulations proposed in the past for steel rods (from simple
braided, and ribbed)-were used as reinforcement; three val- linear to more refined nonlinear models), for FRP rebars an
ues of the embedment length-to-diameter ratio were consid- extensive research effort is still needed; furthermore, the few
ered (10, 20, and 40). available formulations for FRP rebars still have to be validated
by comparison with experimental results.
Effect of Temperature Change In the following, a brief review of the available models for
FRP rebars is reported along with an assessment of their re-
The effect of temperature changes on the pullout resistance liability to simulate the experimental results. It should be em-
of smooth rods was investigated by AI-Zaharani (1995). The phasized that the analytical models of the T-S relationship al-
author carried out an experimental program consisting of pull- ready developed are aimed at identifying a general law, which
out tests at different thermal conditions. In particular, three holds always by determining its parameters by curve fitting.
conditions, characterized by different curing and testing tem- No specific formulations for the different types of rebars have
peratures, but the same absolute difference in temperature (dt been developed so far.
= 40°C) between curing and testing, were considered. The re-
sults showed a much lower bond resistance when reducing the Malvar Model
temperature between curing and testing by dt than when in-
creasing it. This effect is due to bar contraction and expansion, The first modeling of the bond phenomenon in the case of
caused by the temperature reduction and increase, respectively, FRP rods was given by Malvar (1994), who carried out an
since both phenomena significantly influence the friction com- extensive experimental research on GFRP bars characterized
ponent of the bond mechanism. Therefore, besides the surface by different shapes of the outer surface. Tests were performed
roughness, the coefficient of thermal expansion of FRP rods for different values of the confinement pressure and for a fixed
is an important parameter for determining the bond behavior tensile strength of the concrete (tests A, B, C, and D reported
of smooth rods, since it is responsible for a change in radial in Table 2).
pressure on the rod when thermal conditions vary. According to test results, Malvar proposed a refined model
Honma and Maruyama (1989) investigated the effect of of the overall bond behavior, depending on two empirical con-
temperature on CFRP rods, performing tests at room temper- stants (F and G) to be determined by curve-fitting the experi-
ature after maintaining the specimens for 3 hours at different mental T-S curves. This type of model is represented by the
preset temperatures (from +20°C to + 100°C); they found that following relationship:
high temperature has a very detrimental effect on bond, prob-
ably as a result of the lower shear stiffness of the matrix of ..!.. _ F(sls m) + (G - 1)(SISm)2
(1)
FRP rods. 'rm - 1 + (F - 2)(slsm )+ G(SISm )2
I !
it 3
821 83 i i
8 0
FIG. 6. (a) BPE Model; (b) Modified BPE Model
(3)
on curve-fitting of the actual data.
It is possible to observe that the CMR model provides an
where Tl = maximum bond strength; and SI = corresponding initial slope (Le., for slip S = 0) equal to infinity, thus allowing
slip (therefore TI = T m and SI = sm)' the physical phenomenon of adhesion to be reproduced quite
In (3) a is a curve-fitting parameter that must not be larger well. The BPE model features the same property [see (3)],
than 1, to be physically meaningful: the value of a proposed whereas the initial slope provided by the Malvar model has a
by Eligehausen et al. (1983) in the case of steel bars and sub- finite value, equal to FTm/Sm (Le., equal to F times the secant
sequently adopted in the "CEB-FIP Model Code 1990" tangent at the peak bond stress).
(1991) is equal to 0.40.
Furthermore, the BPE model presents a second branch with
constant bond (T =TI) up to a slip S =S2; a linearly descending
ANALYTICAL VERSUS EXPERIMENTAL RESULTS
branch from (S2' TI) to (S3' T3); and a horizontal branch for S Based on the available experimental tests, the main param-
> S3, with a value of T due to the development of friction (T eters of the modified BPE model (a, p, T3) and of the CMR
= T3)' Values of S2, S3' and T3 have to be calibrated on the basis model (13, s,) have been calibrated. Parameters 13 and s, of the
of the experimental results. CMR model [see (5)] have been obtained from the experi-
mental data by the least-square error method, while parameter
BPE Modified Model a of the BPE model [see (3)] has been evaluated by equating
Cosenza et al. (1996) proposed an alternative analytical the area AT1 underneath the ascending branch of the analytical
model obtained by slightly modifying the BPE one. In fact, in bond-slip curve [see Fig. 6(b)] given by
the case of FRP rebars, the comparison between the experi-
mental curves and the analytical ones obtained by applying AT! =
SI T(s)'ds = lSI ()" 'ds = - -
S T 1 'SI
(6)
the original BPE model has shown a lack of the second branch,
thus suggesting not to consider it. Then the modified BPE
l o 0
Tj' -
Sl 1+ a
law-shown in Fig. 6(b)-presents the same ascending to the area A Ts underneath the ascending branch of each actual
branch as the original one [see (3)], a softening branch, having curve.
slope P' T/SI from (SI> TI) to (S3' T3)' which is given by In (6) TI and SI represent the bond strength and the corre-
sponding slip, respectively. Therefore, a can be expressed as
2.. =
TI
1- p (!... -
SI
1) (4)
a function of A TI through
and, finally, for S > S3' a horizontal branch simulating the fric- (7)
tion component T 3.
Therefore, in the case of the modified model, only three TABLE 4. Mean Values and Coefficients of Variation for Theo-
parameters have to be estimated: parameter a, which the as- retical Parameters
cending branch depends on; parameter p, accounting for the
softening; and the friction component T3 of the bond resistance. Modified BPE Model CMR Model
Furthermore, this simplified formulation, which appears more T3 s,
suitable for FRP rebars, provides powerful analytical solutions Type of outer surface Ol p (MPa) 13 (mm)
for different problems in the structural mechanics of concrete (1) (2) (3) (4) (5) (6)
members reinforced with FRP rebars. Smooth (number of tests = 10) 0.145 1.87 0.99 0.314 0.11
(0.44) (0.53) (0.47) (0.90) (0.81)
CMR Model Grain covered (number of tests = 5)
0.067 3.11 3.17 0.138 0.07
(0.81) (0.96) (0.22) (1.50) (0.12)
Most structural problems are to be dealt with at the ser- Sandblasted-type (number of tests 0.251 2.63 1.38 0.559 0.41
viceability state level; therefore, a refined modeling of the = 8) (0.97) (0.47) (0.53) (0.52) (1.81)
T-S curve is needed for the ascending branch only (Le., for slip Twisted (number of tests = 3) 0.175 4.15 3.68 0.593 0.12
less than sm). (0.44) (0.11) (0.15) (0.57) (0.24)
For this reason, a new model for the ascending branch of Ribbed (number of tests = 27) 0.283 14.88 7.79 0.575 0.45
(0.56) (1.20) (0.60) (0.52) (1.60)
the T-S curve has been proposed by Cosenza et al. (1995). This 12.80 6.26 0.473 0.78
Indented and braided (number of 0.177
model represents an alternative to the BPE model [(3)] and is tests = 12) (0.05) (0.61) (0.36) (0.66) (0.76)
defined by the following expression: Braided and sanded (number of 0.069 0.95 7.13 0.025 0.08
tests = 2) (0.07) (0.16) (0.33) (0.60) (0.07)
-TT = (1 - exp{ -s/s,})~ (5) Note: Numbers in parentheses represent the coeffiCient of vanatlOn.
m
cending curves can be well reproduced using a rigid law. are reported in Figs. 7 -10. To better estimate the relative ad-
Moreover, looking at the BPE model, a mean value of ex equal equacy of different models, each analytical curve is obtained
to about 0.23 has been computed for defonned bars (including by using the coefficients calibrated to its experimental coun-
both ribbed and indented ones); then, the comparison between terpart instead of using the mean values of Table 4. In partic-
the foregoing value and the one proposed for defonned steel ular, Fig. 7 shows a comparison between the BPE model, the
bars (ex = 0.40) show that the FRP defonned rebars have a CMR model, and two experimental curves; it is seen that the
stiffer initial bond behavior than the steel ones. best correlation of the test results is obtained by the CMR
18
model, which seems to be the most suitable formulation for
the ascending branch.
16 A comparison of the analytical curves obtained with the
14 modified BPE model and the experimental results is reported
in Figs. 8 and 9 for smooth and deformed rods, respectively.
12
The experimental bond behavior of defonned and smooth bars
10 is quite different: in most cases T-S curves of defonned rebars
are characterized by remarkable softening, whereas, as already
shown, bond behavior of smooth bars is practically rigid-plas-
tic, with s.. being very small and the softening branch almost
4 • TestBC C.M.R. Model flat. In spite of the significant change in behavior, plots of Figs.
• Test VI B.P.E. Model 8 and 9 show a reasonable agreement between numerical and
2 actual values for both types of rebars.
s[mm]
o L - _ - L_ _.l..-_----'-_ _.J......_---'-_ _. . . l . . - _ - - - - I . _ - ' Finally a comparison of the Malvar and the BPE models
o 0.2 0.4 0.6 0.8 1.2 1.4 1.6
with some experimental curves is reported in Figs. lO(a) and
FIG. 7. BPE Model versus CMR Model lO(b).
2.------------------...,
(MPal o Tes/83R Tes/CV6
D
- - B.P.F.. Maciel o Tes/eVJl
1.6 - - B.P.E. Model
1.2
000 0
0.4
2.-------------------,
rt (MPal o Tes/A3R Test GV6
D
1.2
0.8 ..........
0.4
"
(c)
8 [mml 8 [mm)
°0.L----l..--........,2'"-----~3'----...
4'--....:.~5
2 3 4 5
FIG. 8. BPE Model versus Experimental Curve: Straight Rebars
o Te.vIA o Te.YIAH
- - IloP.H. Model I:! - - B.P.f:. Model
o 8
~ [mml ()
3 4 () 4
16
t [MPlI]
Downloaded from ascelibrary.org by University of Leeds on 07/01/18. Copyright ASCE. For personal use only; all rights reserved.
o Test BA o TestH5
- - B.P.E. Model 12 - B.P.E. Model
(d)
s[mml s mm
°O·'-----~---~2----3~-.:.l:.:::.::.1.J4
2 3 4
FIG. 9. BPE Model versus Experimental Curve: Deformed Rebars
Thus, both the Malvar model and the modified BPE model
yield good correlation of the entire experimental bond-slip
curves. If one looks at the ascending branch only, which is of
great interest for many practical applications, the CMR model
c T('.'1tF
• n',ytCE6 appears to be the most reliable; however, rather satisfactory
........ MALVAR M",',I correlations can be also obtained by the BPE model, whereas
- M",I(fi<!d B.P.E. Model
the analytical model by Malvar is not very suitable since it
.................... underestimates the initial stiffness compared to the experi-
mental values.
(a) s[mm)
SUMMARY AND CONCLUDING REMARKS
°O·L----:O,.-\.4:---0::-.8:---..,.1-'::.2--:-1,.-\.8:-....:..;~2
In this paper results from several recent experimental stud-
4.5 ........ , " ,' . ies, which have investigated the bond behavior between FRP
t [MPs] rebars and the concrete matrix, are discussed in order to clarify
some important issues and to provide a rational state of the
art on this subject.
3 o T,wF
From the discussion the following observations emerge.
• TestCE6 FRP smooth rods are inadequate for use in reinforced con-
........ MA LVAR Model
- Modified B.P.E. Model crete structures. In fact, the actual bond-slip curves show very
low values of bond strength due to the activation of a friction
1.5 mechanism with large damage of the rebar surface, without
cracks in the surrounding concrete. Moreover, the bond
strength is seen to be dependent on fiber and resin properties,
(b) whereas it is not affected by concrete strength.
s[mml
°O·L...---=O-=.O=-S-----;O"".1-=S---:-O::-.2::::S:-----..,..;;,:;.;.:~O.35 Sand-covered continuous fiber rebars show good bond re-
sistance; however, the interface between sand grains and bars
FIG. 10. BPE versus Malvar Model: (a) Entire Curve; (b) First detaches abruptly, thus leading to a brittle bond failure.
Branch Deformation of a smooth rod surface obtained by gluing a
spiral on it does not result in any significant improvement in
Fig. IO(a) shows the comparison over the entire curve; bond behavior. In fact, the test results show that the bond
therefore, parameters F and G of the Malvar law have been collapse is due to the detachment of the spiral, while the con-
determined by curve-fitting the two selected experimental crete is uncrushed. Therefore, the bond strength is provided
curves. A good agreement between both the analytical models by a friction mechanism and Tm values are slightly higher than
and the experimental data can be observed. Nevertheless, the those of smooth bars.
examination of Fig. lO(b), which reports a magnification of Twisted strands present slightly larger values of bond
the ascending branch, indicates that a better modeling of the strength T m compared to those obtained for smooth rebars. A
initial behavior is provided by the BPE model, which closely low dependency of bond resistance on the concrete compres-
predicts the initial stiffness of the experimental curves. sive strength has been evidenced by tests.
50 I JOURNAL OF COMPOSITES FOR CONSTRUCTION I MAY 1997
Four analytical fonnulations have been considered to rep- Kanakubo, T., Yonemaru, K., Fukuyama, H., Fujisawa, M., and Sonobe,
resent bond-slip behavior. The Malvar model is able to repro- Y. (1993). "Bond performance of concrete members reinforced with
duce the entire constitutive T-S curve by means of only one FRP bars." Proc., Int. Symp. on Fiber Reinforced Plastic Reinforce-
relation, but it appears less reliable to model the ascending ment for Concrete Struct.; ACI SP-/38, A. Nanni and C. W. Dolan,
eds.
branch than the BPE and the CMR models. The modified BPE Larralde, J., and Silva Rodriguez, R. (1993). "Bond and slip of FRP
model is also suitable for reproduction of the entire curve; rebars in concrete." J. Mat. in Civ. Engrg., ASCE, 5(1).
furthennore, it shows a good agreement with the experimental Makitani, E., Irisawa, I., and Nishiura, N. (1993). "Investigation of bond
data even within the first branch. The CMR model describes in concrete member with fiber reinforced plastic bars." Proc., Int.
only the ascending branch, but, in this range, it is proved to Symp. on Fiber-Reinforced-Plastic Reinforcement for Concrete Stmct.;
lead to the best simulations. ACI SP-138, A. Nanni and C. W. Dolan, eds.
Malvar, L. J. (1994). "Bond stress-slip characteristics of FRP rebars."
Rep. TR-2013-SHR, Naval Fac. Engrg. Service Ctr., Port Hueneme,
APPENDIX I. REFERENCES Calif.
Mashima, M., and Iwamoto, K. (1993). "Bond characteristics of FRP rod
AI-Dulaijan, S. U., Nanni, A., AI-Zaharani, M. M., Bakis, C. E., and and concrete after freezing and thawing deterioration." Proc., Int.
BoothbY, T. E. (1996). "Bond evaluation of environmentally condi- Symp. on Fiber-Reinforced-Plastic Reinforcement for Concrete Struct.;
tioned GFRP/concrete systems." Proc., 2nd Int. Cant on Advanced ACI SP-138, A. Nanni and C. W. Dolan, eds.
Compos. Mat. in Bridge Struct., M. EI-Badry, ed. Nanni, A.• AI-Zaharani, M. M., AI-Dulaijan. S. U., Bakis, C. E., and
Alunno Rossetti, V., Galeota, D., and Giammatteo, M. M. (1995). "Local Boothby, T. E. (1995). "Bond of FRP reinforcement to concrete-
bond stress-slip relationships of glass fibre reinforced plastic bars em- experimental results." Proc., 2nd Int. RILEM Symp. (FRPRCS-2J, L.
bedded in concrete." Mat. and Struct., 28(180). Taerwe, ed.
AI-Zaharani, M. M. (1995). "Bond behaviour of fiber reinforced plastic Soroushian, P., Choi, K. B., Park, G. H., and Aslani, F. (1991). "Bond
(FRP) reinforcements with concrete," PhD thesis, Pennsylvania State of deformed bars to concrete: effects of confinement and strength of
Univ., University Park, Pa. concrete." ACI Mat. J., 88(3).
AI-Zaharani, M. M., Nanni, A., AI-Dulaijan, S. U., and Bakis, C. E.
(1996). "Bond of FRP to concrete in reinforcement rods with axisym- APPENDIX II. NOTATION
metric deformations." Proc., 2nd Int. Cant on Advanced Compos. Mat.
in Bridge Struct., M. EI-Badry, ed. The following symbols are used in this paper;
Benmokrane, B., and Masmoudi, R. (1996). "FRP C-bar as reinforcing
rod for concrete structures." Proc.• 2nd Int. Cant on Advanced Com- A,B,C,D,E empirical constants of Eqs. (2);
pos. Mat. in Bridge Struct., M. EI-Badry, ed. F,G empirical constants of Malvar model;
Benmokrane. B., Tighiouart, B., and Chaallal, O. (1996). "Bond strength
and load distribution of composite GFRP reinforcing bars in concrete."
fc compressive concrete strength;
f, tensile concrete strength;
ACI Mat. J., 93(3).
Chaallal, 0., and Benmokrane, B. (1993). "Pullout and bond of glass-
p parameter of BPE modified model [Eq. (4)];
S = slip;
fibre rods embedded in concrete and cement grout." Mat. and Stmct.,
26. Sm slip at maximum bond stress;
Comite Euro-International du Beton (CEB). (1991). "CEB-PIP model S, = slip at end of BPE model ascending branch;
code 1990." Bull. d'information, Lausanne, SWitzerland, 203-205. S2 slip at beginning of BPE model softening
Cosenza, E., Manfredi, G., and Realfonzo, R. (1995). "Analytical mod- branch;
elling of bond between FRP reinforcing bars and concrete." Proc., 2nd S3 slip at beginning of BPE model frictional
Int. RILEM Symp. (FRPRCS-2J, L. Taerwe, ed. branch;
Cosenza, E., Manfredi, G., and Realfonzo, R. (1996). "Bond character- <X parameter of BPE model [(Eq. 3)];
istics and anchorage length of FRP rebars." Proc., 2nd Int. Cant on
~, Sr = parameters of CMR model [(Eq. 5)];
Advanced Compos. Mat. in Bridge Struct., M. EI-Badry, ed.
CI = confining axisymmetric radial pressure;
Daniali, S. (1992). "Development length for fibre-reinforced plastic
bars." Proc., 1st Int. Cont on Advanced Compos. Mat. in Bridge T bond stress;
Struct.• K. W. Neale and P. Labossi~re, eds. Tm maximum bond stress;
Ehsani, M. R., Saadatrnanesh, H., and Tao, S. (1993). "Bond of GFRP TI = maximum bond stress of BPE model; and
rebars to ordinary-strength concrete." Proc.• Int. Symp. on Fiber Re- T3 = frictional component of BPE model bond.